MNRAS 444, 3427–3445 (2014) doi:10.1093/mnras/stu570 The ATLAS3D Project – XXVIII. Dynamically driven star formation suppression in early-type galaxies Timothy A. Davis,1‹ Lisa M. Young,2,3 Alison F. Crocker,4 Martin Bureau,5 Leo Blitz,6 Katherine Alatalo,7 Eric Emsellem,1,8 Thorsten Naab,9 Estelle Bayet,5 Maxime Bois,10 Fred´ eric´ Bournaud,11 Michele Cappellari,5 Roger L. Davies,5 P. T. de Zeeuw,1,12 Pierre-Alain Duc,11 Sadegh Khochfar,13 Davor Krajnovic,´ 14 Harald Kuntschner,15 Richard M. McDermid,16 Raffaella Morganti,17,18 17,18 19 20 17,21 Tom Oosterloo, Marc Sarzi, Nicholas Scott, Paolo Serra Downloaded from and Anne-Marie Weijmans22 Affiliations are listed at the end of the paper Accepted 2014 March 19. Received 2014 March 18; in original form 2013 December 18 http://mnras.oxfordjournals.org/ ABSTRACT We present measurements of the star formation rate (SFR) in the early-type galaxies (ETGs) of the ATLAS3D sample, based on Wide-field Infrared Survey Explorer (WISE)22μm and Galaxy Evolution Explorer far-ultraviolet emission. We combine these with gas masses estimated 12 from CO and H I data in order to investigate the star formation efficiency (SFE) in a larger sample of ETGs than previously available. We first recalibrate (based on WISE data) the at European Southern Observatory on October 27, 2015 relation between old stellar populations (traced at Ks band) and 22 μm luminosity, allowing us to remove the contribution of 22 μm emission from circumstellar dust. We then go on to investigate the position of ETGs on the Kennicutt–Schmidt (KS) relation. Molecular gas-rich ETGs have comparable star formation surface densities to normal spiral galaxy centres, but they lie systematically offset from the KS relation, having lower SFEs by a factor of ≈2.5 (in agreement with other authors). This effect is driven by galaxies where a substantial fraction of the molecular material is in the rising part of the rotation curve, and shear is high. We show here for the first time that although the number of stars formed per unit gas mass per unit time is lower in ETGs, it seems that the amount of stars formed per free-fall time is approximately constant. The scatter around this dynamical relation still correlates with galaxy properties such as the shape of the potential in the inner regions. This leads us to suggest that dynamical properties (such as shear or the global stability of the gas) may be important second parameters that regulate star formation and cause much of the scatter around star formation relations. Key words: stars: mass-loss – ISM: molecules – galaxies: elliptical and lenticular, cD – galaxies: ISM. be an artefact of the imperfect methods we have of estimating star 1 INTRODUCTION formation rates (SFRs), and tracing molecular hydrogen (Genzel Star formation (SF) is a fundamental process, responsible for con- et al. 2012). verting the soup of primordial elements present after the big bang Atomic gas is present in ≈32 per cent of early-type galax- into the Universe we see around us today. Despite this, debate still ies (ETGs; Bottinelli & Gouguenheim 1977; Knapp, Turner & rages about the way SF proceeds, and the role (if any) that environ- Cunniffe 1985; Morganti et al. 2006; di Serego Alighieri et al. 2007; ment plays in its regulation. For instance, high-redshift starbursts Grossi et al. 2009; Oosterloo et al. 2010; Serra et al. 2012), dust seem to convert gas into stars much more efficiently than local disc in ≈60 per cent (Colbert, Mulchaey & Zabludoff 2001; Smith et al. galaxies (Daddi et al. 2010; Genzel et al. 2010). This increased 2012; Agius et al. 2013), and molecular gas in 22 per cent (Combes, efficiency may be explained by a change in gas properties (e.g. the Young & Bureau 2007; Welch, Sage & Young 2010; Young et al. high fraction of gas at high volume densities in starbursts), or may 2011, hereafter Paper IV). Low-level residual SF has also been de- tected through studies of ultraviolet (UV) emission (e.g. Yi et al. E-mail: [email protected] 2005; Kaviraj et al. 2007; Salim & Rich 2010; Wei et al. 2010), C 2014 The Authors Published by Oxford University Press on behalf of the Royal Astronomical Society 3428 T. A. Davis et al. optical emission lines (e.g. Crocker et al. 2011) and infrared emis- 2.1 Molecular gas masses sion (e.g. Knapp et al. 1989; Combes et al. 2007; Temi, Brighenti The CO(1-0) and CO(2-1) lines were observed in every galaxy in the & Mathews 2009, hereafter T09; Shapiro et al. 2010). ATLAS3D sample at the IRAM 30 m telescope, and 56 objects were Typically ETGs have much smaller fraction of molecular gas to detected (for full details see Paper IV). From these observations we stellar mass than spirals. This average fraction appears to decrease have estimated molecular gas masses for the detected galaxies, using with increasing galaxy bulge fraction (Cappellari et al. 2013b, here- a Galactic X factor of 3 × 1020 cm−2 (K km s−1)−1 (Dickman, after Paper XX; see also Saintonge et al. 2012). This suggests CO Snell & Schloerb 1986). We return to discuss this assumption later, a connection between bulge formation and galaxy quenching, as but as ETGs usually have high stellar metallicities such a value is also suggested by optical studies (Bell et al. 2012). However the a priori reasonable. Making this assumption, we found molecular decrease of the molecular gas fraction does not seem to be the gas reservoirs with masses between 106 and 109.5 M, as tabulated only factor making ETGs red. In fact, even at fixed gas fraction, in Paper IV. We were also able to place limits on the amount of molecule-rich ETGs form stars less efficiently than normal spirals, molecular gas of CO non-detected objects, finding upper limits and very much less efficiently than high-redshift starburst galaxies between 106 and 108 M (for objects at different distances). (Saintonge et al. 2011, 2012; Martig et al. 2013, hereafter Paper These observations were single pointings at the galaxy centres, XXII). Such a suppression would help explain how objects in the with a beam size of ≈22 arcsec for the CO(1-0) transition (used to Downloaded from red sequence can harbour substantial cold gas reservoirs for a long calculate the molecular gas masses). In some objects the molecular period of time, without becoming significantly blue. A similar sup- gas distribution was later shown to be more extended than the 30 m pression of SF may also be ongoing in the central parts of our own telescope beam (see Davis et al. 2013a, hereafter Paper XIV,foran Milky Way (Longmore et al. 2013), suggesting this may be a general analysis of the total molecular gas extent in these objects). In these process in spheroids and/or dense stellar environments. The physics cases, we use total interferometric CO fluxes from Alatalo et al. of whatever process is causing this suppression of SF is, however, http://mnras.oxfordjournals.org/ (2013, hereafter Paper XVIII). In principle it is possible that these unknown. The deep potential wells of these objects could hold gas interferometric observations resolved out some emission, which stable against collapse (dubbed ‘morphological quenching’; Martig would make our CO masses lower limits. The correction for molec- et al. 2009), or strong tidal fields and streaming motions could pull ular material outside the beam of our single-dish observations is clouds apart (e.g. Kruijssen et al. 2013; Meidt et al. 2013), lowering much more significant, however, and so we consider it better to use the observed star formation efficiency (SFE). the interferometric fluxes where possible. As the CO is not generally In this work we use data from the ATLAS3D project to investi- extremely extended, we do not expect the amount of flux resolved gate if local ETGs do display a lower SFE than local spirals, and if out to be large, so this should not affect our conclusions. In objects so what may be driving this suppression. ATLAS3D is a complete, < without interferometric observations, we used the single-dish CO volume-limited exploration of local ( 42 mpc) ETGs (Cappellari at European Southern Observatory on October 27, 2015 fluxes to estimate the masses. Our size estimates (described below) et al. 2011a, hereafter Paper I). All 260 ATLAS3D sample galaxies suggest that very few of these unmapped objects have extended gas have measured total molecular gas masses (or upper limits; from reservoirs, so these 30 m telescope measurements are unlikely to Institut de Radioastronomie Millimetrique´ (IRAM) 30 m CO ob- miss substantial amounts of molecular material. servations presented in Paper IV). H I masses are also available for the northern targets (from Westerbork Synthesis Radio Telescope, WSRT, observations; Serra et al. 2012, hereafter Paper XIII). To es- 2.2 Atomic gas masses timate the SFR in these objects, we utilize data from the Wide-field As presented in Paper XIII, all ATLAS3D field galaxies above a Infrared Survey Explorer (WISE; Wright et al. 2010) all sky survey declination of 10◦ were observed with the WSRT, with a resolution μ at 22 m, and from the Galaxy Evolution Explorer (GALEX)inthe of ≈35 arcsec. For Virgo cluster objects we take the data from the far-ultraviolet (FUV). arecibo legacy fast ALFA (ALFALFA) survey (di Serego Alighieri Section 2 presents the data we use in this work, and describes et al. 2007), as documented in Paper XIII. Most of the molecular how derived quantities are calculated. Section 3 presents our results, discs studied here are smaller than 35 arcsec, so we assume that μ where we investigate the 22 m emission from CO non-detected only the H I gas mass detected in the innermost beam is important.